feat: add/update schemas, config, README, and test

README.md

@@ -13,3 +13,15 @@ This runner is expected to be implemented for a water data hub with the relevant
 ```bash
 cargo run --release
 ```
+
+## Run tests
+
+```bash
+cargo test
+```
+
+To include print statements in test output, run:
+
+```bash
+cargo test -- --nocapture
+```

config.toml

@@ -1,2 +0,0 @@
-ckan_url = "https://data.dathere.com"
-geoconnex_about_field_name = "geoconnex_about"

config/example.toml

@@ -0,0 +1,9 @@
+# Example configuration file. Customize for a specific CKAN instance when making a new Docker image.
+
+# This is the path after https://geoconnex.us/ and before /{id}.
+# Same as namespace location at https://github.com/internetofwater/geoconnex.us.
+namespace_path = "nmwdh/ckan-dataset"
+# The base URL for the CKAN instance to run API requests to.
+ckan_url = "http://localhost:5000"
+# The name of the key in /package_show for a given CKAN dataset where the Geoconnex `about` field is populated.
+about_field_name = "geoconnex_about"

src/lib.rs

@@ -0,0 +1,45 @@
+use serde_json::json;
+
+pub fn get_dataset_schema() -> serde_json::Value {
+    json!({
+        "type": "object",
+        "properties": {
+            "@context": {"type": ["string", "object"]},
+            "@type": {"type": ["string", "array"], "contains": {"const": "Dataset"}},
+            "@id": {"type": "string"},
+            "name": {"type": "string"},
+            "provider": {
+                "type": "object",
+                "properties": {"@type": {"type": "string"}, "name": {"type": "string"}},
+            },
+            "about": {
+                "type": ["string", "array"],
+                "items": {
+                    "type": "object",
+                    "properties": {"@id": {"type": "string"}, "@type": {"const": "Place"}},
+                },
+                "minItems": 1
+            },
+        },
+        "required": ["@context", "@type", "@id", "name", "provider", "about"]
+    })
+}
+
+pub fn get_location_schema() -> serde_json::Value {
+    json!({
+        "type": "object",
+        "properties": {
+            "@context": {"type": ["string", "object"]},
+            "@type": {"type": ["string", "array"], "contains": {"const": "Place"}},
+            "@id": {"type": "string"},
+            "name": {"type": "string"},
+            "provider": {
+                "type": "object",
+                "properties": {"@type": {"type": "string"}, "name": {"type": "string"}},
+            },
+            "geo": {"type": "object"},
+            "gsp:hasGeometry": {"type": "object"}
+        },
+        "required": ["@context", "@type", "@id", "name", "provider", "geo", "gsp:hasGeometry"]
+    })
+}

src/main.rs

@@ -1,29 +1,21 @@
-mod utils;
-
 use anyhow::Result;
 
 #[tokio::main]
 async fn main() -> Result<(), Box<dyn std::error::Error>> {
     let ckan = ckanaction::CKAN::builder()
-        .url("https://data.dathere.com")
+        .url("http://localhost:5000")
         .build();
-    // TODO: Paginate through package_list and run package_show for each package
-    // If about exists then construct JSON-LD and validate then output JSON-LD to stdout on a new line
-    let response = ckan.package_show().id("".to_string()).call().await?;
+
+    // Paginate through /api/3/action/package_list until only an empty array is returned
+    let response = ckan.package_list().call().await?;
 
     let result = response
         .as_object()
         .unwrap()
         .get("result")
         .unwrap()
-        .as_object()
+        .as_array()
         .unwrap();
-    if let Some(geoconnex_about) = result.get("geoconnex_about") {
-        // Check if at least one valid reference feature exists in dataset metadata
-    }
-    // TODO: Construct JSON-LD if valid `about`
-    // TODO: Validate constructed JSON-LD against JSON schema
-    // TODO: Print JSON-LD to new line
 
     println!("{result:#?}");
 

src/utils.rs

@@ -1,14 +0,0 @@
-use serde_json::json;
-
-async fn construct_jsonld(metadata: serde_json::Value) -> serde_json::Value {}
-
-async fn get_dataset_schema() -> serde_json::Value {
-    json!({
-        "type": "object",
-        "properties": {
-            "@context": {"type": ["string", "object"]},
-            "@type": {"const": "Dataset"}
-        },
-        "required": []
-    })
-}

tests/validate_jsonld.rs

@@ -0,0 +1,58 @@
+use anyhow::{Result, bail};
+use serde_json::json;
+
+#[test]
+fn validate_usgs_location_jsonld() -> Result<()> {
+    let usgs_location_jsonld = json!({
+      "@context": {
+        "@vocab": "https://schema.org/",
+        "gsp": "http://www.opengis.net/ont/geosparql#",
+        "hyf": "https://www.opengis.net/def/schema/hy_features/hyf/",
+        "locType": "https://api.waterdata.usgs.gov/ogcapi/v0/collections/site-types/items/"
+      },
+      "@type": [
+        "Place",
+        "hyf:HY_HydrometricFeature",
+        "hyf:HY_HydroLocation",
+        "locType:ST-CA"
+      ],
+      "@id": "https://geoconnex.us/usgs/monitoring-location/USGS-253937080285200",
+      "name": "BLACKCREEKCANALWESTOFSOUTHMIAMI FLA",
+      "identifier": {
+        "@type": "PropertyValue",
+        "propertyID": "USGS site identifier",
+        "value": "253937080285200"
+      },
+      "url": "https://api.waterdata.usgs.gov/ogcapi/v0/collections/monitoring-locations/items/USGS-253937080285200",
+      "provider": {
+        "@type": "GovernmentOrganization",
+        "name": "U.S. Geological Survey"
+      },
+      "geo": {
+        "@type": "GeoCoordinates",
+        "latitude": 25.6606597832648,
+        "longitude": -80.4808896071386
+      },
+      "gsp:hasGeometry": {
+        "@type": "http://www.opengis.net/ont/sf#Point",
+        "gsp:asWKT": {
+          "@type": "gsp:wktLiteral",
+          "@value": "POINT (-80.4808896071386 25.6606597832648)"
+        },
+        "gsp:crs": {
+          "@id": "http://www.opengis.net/def/crs/OGC/1.3/CRS84"
+        }
+      }
+    });
+
+    let dataset_json_schema = ckan_geoconnex_bulk_runner::get_location_schema();
+
+    if let Err(e) = jsonschema::validate(&dataset_json_schema, &usgs_location_jsonld) {
+        println!("Error during validation:");
+        bail!("{e}");
+    } else {
+        println!("Successfully validated.");
+    }
+
+    Ok(())
+}